Three-dimensional data encoding method, three-dimensional data decoding method, three-dimensional data encoding device, and three-dimensional data decoding device
Abstract
A three-dimensional data encoding method includes: generating quantized values by quantizing, using one or more first parameters for one or more groups to which one or more attribute information belong, a coefficient value calculated for each node included in a layered structure of three-dimensional points. At least one attribute information belongs to any of the one or more groups. The calculating includes calculating, using two first coefficient values of two first nodes of a first layer, a second coefficient value of a second node of a second layer which is one layer higher. The quantizing includes: shifting-up the one or more first parameters by N bits (N is a natural number); shifting-down, by one bit, a sum of two second parameters of the two first nodes; shifting-down the result by N bits; and quantizing the second coefficient value using the obtained parameter.
Claims
exact text as granted — not AI-modifiedWhat is claimed is:
1. A three-dimensional data encoding method comprising:
calculating a coefficient value of each of nodes included in a layered structure by generating the layered structure in which each of items of attribute information of three-dimensional points included in point cloud data is hierarchically divided into a first component and a second component; and
quantizing, for each of the nodes, the coefficient value of the node using one or more first parameters for one or more groups to which one or more items of attribute information used in the calculating of the coefficient value belong, to generate quantized values, wherein
at least one item of attribute information among the items of attribute belongs to any one of the one or more groups,
in the calculating, a second coefficient value of a second node of a second layer which is one layer higher than a first layer of the layered structure is calculated using two first coefficient values of two first nodes belonging to the first layer,
the quantizing includes:
(i) generating one or more second parameters by left-shifting the one or more first parameters by N bits, N being a natural number;
(ii) generating a third parameter by right-shifting, by one bit, a sum obtained by adding up two second parameters from among the one or more second parameters, the two second parameters corresponding to the two first nodes;
(iii) generating a fourth parameter by right-shifting the third parameter by N bits; and
(iv) quantizing the second coefficient value using the fourth parameter, and
each parameter of the one or more first parameters, the one or more second parameters, the third parameter, and the fourth parameter is expressed by a bit sequence.
2. The three-dimensional data encoding method according to claim 1 , wherein
the first parameter and the fourth parameter are each differences from a reference parameter, and
in the quantizing of the second coefficient value, a fifth parameter obtained by adding the reference parameter to the fourth parameter is used.
3. The three-dimensional data encoding method according to claim 1 , wherein N is 4.
4. The three-dimensional data encoding method according to claim 1 , wherein
each of the items of attribute information is classified into any one of the one or more groups, according to a three-dimensional space to which the three-dimensional point having the item of attribute information belongs.
5. A three-dimensional data decoding method comprising:
calculating, using quantized values and one or more first parameters, coefficient values by inverse-quantizing each of the quantized values; and
calculating, from the coefficient values, items of attribute information of three-dimensional points included in point cloud data, wherein
the one or more first parameters are one or more parameters for one or more groups to which one or more items of attribute information belong, the one or more items of attribute information being used in the calculating of a coefficient value to which a corresponding one of the quantized values corresponds,
at least one item of attribute information among the items of attribute information belongs to any one of the one or more groups,
the coefficient values are calculated corresponding to nodes included to a layered structure,
the layered structure is a structure in which each of the items of attribute information is hierarchically divided into a first component and a second component,
two first coefficient values of two first nodes belonging to a first layer of the layered structure are used in calculating a second coefficient value of a second node of a second layer which is one layer higher than the first layer,
the inverse-quantizing includes:
(i) generating one or more second parameters by left-shifting the one or more first parameters by N bits, N being a natural number;
(ii) generating a third parameter by right-shifting, by one bit, a sum obtained by adding up two second parameters from among the one or more second parameters, the two second parameters corresponding to the two first nodes;
(iii) generating a fourth parameter by right-shifting the third parameter by N bits; and
(iv) inverse-quantizing a quantized value of the second node using the fourth parameter, and
each parameter of the one or more first parameters, the one or more second parameters, the third parameter, and the fourth parameter is expressed by a bit sequence.
6. The three-dimensional data decoding method according to claim 5 , wherein
the first parameter and the fourth parameter are each differences from a reference parameter, and
in the inverse-quantizing of the quantized value of the second node, a fifth parameter obtained by adding the reference parameter to the fourth parameter is used.
7. The three-dimensional data decoding method according to claim 5 , wherein N is 4.
8. The three-dimensional data decoding method according to claim 5 , wherein
each of the items of attribute information is classified into any one of the one or more groups, according to a three-dimensional space to which the three-dimensional point having the item of attribute information belongs.
9. A three-dimensional data encoding device comprising:
a processor; and
memory, wherein
using the memory, the processor:
calculates a coefficient value of each of nodes included in a layered structure by generating the layered structure in which each of items of attribute information of three-dimensional points included in point cloud data is hierarchically divided into a first component and a second component; and
quantizes, for each of the nodes, the coefficient value of the node using one or more first parameters for one or more groups to which one or more items of attribute information used in the calculating of the coefficient value belong, to generate quantized values,
at least one item of attribute information among the items of attribute belongs to any one of the one or more groups,
in the calculating, the processor calculates a second coefficient value of a second node of a second layer which is one layer higher than a first layer of the layered structure, using two first coefficient values of two first nodes belonging to the first layer,
in the quantizing, the processor:
(i) generates one or more second parameters by left-shifting the one or more first parameters by N bits, N being a natural number;
(ii) generates a third parameter by right-shifting, by one bit, a sum obtained by adding up two second parameters from among the one or more second parameters, the two second parameters corresponding to the two first nodes;
(iii) generates a fourth parameter by right-shifting the third parameter by N bits; and
(iv) quantizes the second coefficient value using the fourth parameter, and
each parameter of the one or more first parameters, the one or more second parameters, the third parameter, and the fourth parameter is expressed by a bit sequence.
10. A three-dimensional data decoding device comprising:
a processor; and
memory, wherein
using the memory, the processor:
calculates, using quantized values and one or more first parameters, coefficient values by inverse-quantizing each of the quantized values; and
calculates, from the coefficient values, items of attribute information of three-dimensional points included in point cloud data,
the one or more first parameters are one or more parameters for one or more groups to which one or more items of attribute information belong, the one or more items of attribute information being used in the calculating of a coefficient value to which a corresponding one of the quantized values corresponds,
at least one item of attribute information among the items of attribute information belongs to any one of the one or more groups,
the coefficient values are calculated corresponding to nodes included to a layered structure,
the layered structure is a structure in which each of the items of attribute information is hierarchically divided into a first component and a second component,
two first coefficient values of two first nodes belonging to a first layer of the layered structure are used in calculating a second coefficient value of a second node of a second layer which is one layer higher than the first layer,
in the inverse-quantizing, the processor:
(i) generates one or more second parameters by left-shifting the one or more first parameters by N bits, N being a natural number;
(ii) generates a third parameter by right-shifting, by one bit, a sum obtained by adding up two second parameters from among the one or more second parameters, the two second parameters corresponding to the two first nodes;
(iii) generates a fourth parameter by right-shifting the third parameter by N bits; and
(iv) inverse-quantizes a quantized value of the second node using the fourth parameter, and
each parameter of the one or more first parameters, the one or more second parameters, the third parameter, and the fourth parameter is expressed by a bit sequence.Join the waitlist — get patent alerts
Track US12437447B2 — get alerts on status changes and closely related new filings.
We store only your email — no account needed. See our privacy policy.